Latest Articles for Robustness

New methods enhance video action detection against various corruptions.

This study focuses on improving adaptability in deep reinforcement learning algorithms.

A novel approach enhances language model reliability through self-healing mechanisms.

Exploring how transformers favor low sensitivity functions for improved performance.

This article explores PID control integration into transformers to improve robustness and output quality.

Enhancing tools to detect harmful language in online spaces is crucial for safety.

A new method for devices to communicate using on-device AI systems.

Enhancing code generation models to handle minor input changes effectively.

A guide to metrics for assessing LLM performance in various fields.

Measuring robustness and safety in cyber-physical systems under sensor threats.

New approach improves robot object manipulation by measuring energy margins for robustness.

A new method guarantees the validity of counterfactual explanations despite model changes.

New methods improve watermark identification and protection for AI images.

Study reveals how models respond to gibberish prompts and implications for safety.

A method to increase classifier reliability against data manipulation.

A new framework for assessing the performance of RAG systems.

Learn how to create reliable programs that react to their environment.

DiffuseLoco improves robot walking, enabling quick adaptation using past experiences.

New techniques improve model performance against data shifts using non-convex loss functions.

M3Net enhances LiDAR segmentation for self-driving cars by integrating diverse datasets and sensors.

A look into scheduling jobs without knowing their sizes.

New watermarking technique improves protection for AI-generated images.

A new method enhances image recognition by mimicking human visual adjustments.

Assessing the capabilities and challenges of advanced video understanding models.

New algorithms enhance decision-making by considering risk and efficiency.

A new method groups contingencies to enhance power system stability and efficiency.

Engineers use the redundancy matrix to ensure safe and strong structures.

An overview of object detection technology, focusing on model performance and recent advancements.

A method to improve regression model reliability in the face of attacks.

A new method enhances the reliability of GCNs against adversarial changes.

Learn how to analyze changes in outcomes with a robust method.

New method improves neural networks' performance against adversarial attacks.

AI professionals share their views on Trustworthy AI and its challenges.

Examining the role of watermarking in AI-generated text for accountability.

A look at degree correlation and rewiring techniques in networks.

SOBER improves efficiency in finding optimal parameters for complex models.

A new approach to watermarking aims to enhance detection and prevent misuse of language models.

Improving performance and stability in complex control systems.

New preconditioner improves fluid simulations in porous materials for various industries.

Improving machine learning robustness against adversarial examples is critical for secure applications.